Elimination method and system of remaining echoes

Abstract

The invention relates to the field of voice communication, and discloses an elimination method and system of remaining echoes. Windowing processing is carried out on the remaining echoes by adopting a Hamming window or a Blackman window or other non-rectangular window, spectrum leakage can be reduced so that accuracy of linear prediction can be improved, residual error signals are made to be more similar to white noise, and therefore the remaining echoes can be eliminated better. Further, when the windowing processing is carried out on the residual echo signals, remaining echo signals of a base frame and the last N signals of a last frame can be combined to form the remaining echo signals needing to be process at present, and the spectrum leakage can be further reduced.

Description

technical field [0001] The invention relates to the field of voice communication, in particular to a residual echo cancellation method and system. Background technique [0002] In a voice communication system, after the remote input signal arrives at the local signal receiving device, it is processed by the local signal receiving device and reaches the earpiece. During this process, echoes are often generated. In echo cancellation, four processing modules are involved, which are adaptive filtering module, double-ended detection module, residual echo cancellation module, and noise cancellation module. The residual echo cancellation module and the noise cancellation module can be collectively referred to as the nonlinear processing part. Such as figure 1 shown. Among them, the adaptive filter is mainly used to track the echo reflection path and estimate the estimated echo signal The double-talk detector is used to control the update of the adaptive filter to prevent the fi...

AI Technical Summary

Problems solved by technology

[0007] The existing technical solution uses the AR model for whitening residual echo processing. Due to the limitation of system processing capacity, the voice signal needs to be truncated before processing. However, the direct truncation of the voice signal will inevitably cause spectrum leakage and whitening. The effect is not good, and if such a signal is sent to the noise suppressor, the elimination effect will be compromised

Images